Difference between revisions of "Idetifiable Intelligence"
(→Problems) |
(→Problems) |
||
| Line 8: | Line 8: | ||
==Problems== | ==Problems== | ||
| + | AI degrades over time. Perhaps that tells us what happens to humans in a technological world. We need to scrap both as they get poisoned by the real world. | ||
===Model Drift=== | ===Model Drift=== | ||
Model drift occurs when a machine learning model's performance deteriorates over time due to changes in input data distributions, system environments, or user behavior. Drift types include: | Model drift occurs when a machine learning model's performance deteriorates over time due to changes in input data distributions, system environments, or user behavior. Drift types include: | ||
Revision as of 21:54, 26 May 2025
Identifiable Intelligence here means the ability to isolate an instance of an entity with a continuing Identifier and Behavior over time
Context
Nearly all of the first couple of generations of Artificial Intelligence researchers were focused on machines that could emulate human intelligence. The first real breakthroughs were Large Language Models (LLM) that operated using existing computer technologies. The resulting language models could be used in General Purpose Transforms to create responses in language inputs. The model was identifiable, but the particular instance was ephemeral and so had not enduring Identity.
Interfaces to Humans
The question arises from time to time about how humans can interface with Artificial Intelligence.In 2023 when AI began to be useful the technology was completely different from Human Intelligence and the only available interface was the laptop or mobile computer. Ray Kurzweil was one of several futurists who imagined the time when human intelligence could directly interface to mechanical intelligence.[1] One way was Molecular nanotechnology (MNT), a technology based on the ability to build structures to complex, atomic specifications by means of mechanosynthesis. This is distinct from nanoscale materials.
Aims
You awake one morning to find your brain has another lobe functioning. Invisible, this auxiliary lobe answers your questions with information beyond the realm of your own memory, suggests plausible courses of action, and asks questions that help bring out relevant facts. You quickly come to rely on the new lobe so much that you stop wondering how it works. You just use it. This is the dream of artificial intelligence. — Byte, April 1985[2]
Problems
AI degrades over time. Perhaps that tells us what happens to humans in a technological world. We need to scrap both as they get poisoned by the real world.
Model Drift
Model drift occurs when a machine learning model's performance deteriorates over time due to changes in input data distributions, system environments, or user behavior. Drift types include:
- Covariate Drift: Changes in the distribution of input features
- Concept Drift: Changes in the relationship between inputs and outputs
- Prior Probability Drift: Changes in label distribution
These drifts require retraining, patching, or deploying monitoring systems — all of which consume time and resources.
References
- ↑ Ray Kurzweil, The Singularity is Near p. 324ff.
- ↑ Wikipedia Transhumanism https://en.wikipedia.org/wiki/Transhumanism
